The Numbers Come First
Henrik Landgren was Spotify's first-ever VP of Analytics, leading the company's data strategy under CEO Daniel Ek at a time when music streaming had not yet launched in the United States. His task was straightforward and unglamorous: move the company off static Excel spreadsheets and toward dynamic, real-time data tracking using Hadoop — software he describes as haunting data analysts 'to this day.'
The payoff was real. By learning to track where users clicked and where they lingered, Spotify's platform surpassed 20 million subscribers with a 'level of precision that guesswork alone could never have gotten us to,' Landgren writes in a commentary published by Fortune.
What Landgren Found When He Moved to VC
When Landgren carried that discipline into venture capital half a decade later — during Stockholm's post-Klarna, post-Spotify moment as a hotbed of unicorns per capita — he found the investing world largely untouched by data culture. Dinner conversations revolved around 'a charming visionary,' nods and anecdotes generated hype, and hype generated increasingly large funding rounds. The company's actual figures, he writes, barely entered the room.
The consequences are measurable. According to Landgren, nearly two-thirds of venture deals never return the original investment. Yet most firms store their post-mortems 'somewhere nobody typically looks, dusted off only when the next fund's LP needs a slide on lessons learned.' Bessemer Venture Partners is a rare exception, maintaining a public 'anti-portfolio' of famous passes — including Apple's pre-IPO stock, which one Bessemer partner reportedly dismissed at the time as 'outrageously expensive.'
Motherbrain and the Infrastructure Gap
Landgren's response was to build a proprietary tool at his first VC firm. Named Motherbrain — after the Nintendo character — it used signals like traction data, usage numbers, and spikes in web traffic to surface opportunities before competing investors knew to look. Within a few years, colleagues attributed over $100 million of the fund's investments to deals with 'Motherbrain's fingerprints on them.'
The exercise exposed a second problem: when Landgren's team pushed portfolio companies to track their own performance more rigorously, they found that 'nearly none of them had the infrastructure to capture it in the first place.' The AI wave has not solved this. Landgren's assessment is blunt — startups are 'bolting LLMs onto infrastructure never built to feed one,' and VCs are doing the same. His analogy: 'it's like pouring cooking oil in a gasoline car.'
The Editorial Read
Landgren's argument is, at its core, a free-enterprise argument: capital misallocated on vibes rather than evidence is capital destroyed. When two-thirds of VC deals fail to return principal, the losers are not abstract — they are pension funds, university endowments, and individual investors who trusted professionals to do the analytical work.
The deeper problem Landgren identifies is structural. Data teams 'sit outside the main team' and lack the standing to change how a deal actually goes. That is an organizational failure, not a technology gap. AI tools layered on top of that failure do not fix it; they accelerate it. The lesson from Spotify's early days remains the same one the market has always rewarded: discipline before deployment, evidence before conviction, and infrastructure before the intelligence you plan to run on top of it.
